Reduced-order modeling methods of electromagnetic wave propagation in magnetized plasmas

摘要

In this paper, reduced-order modeling methods of electromagnetic wave propagation in magnetized plasmas are considered. Firstly, the Current Density Convolution Finite Difference Time Domain (JEC-FDTD) method is used to obtain the true solution of the scattered electric field in parallel and perpendicular to the magnetic field in plasmas. Secondly, a low-order pattern library based on the dynamic behavior of control system is constructed by Proper Orthogonal Decomposition (POD) and machine learning. Finally, the model which is combined with compressive and POD library is verified by constructing the temporal-spatial electric fields with few samplings. The model will be of great significance for the invertion of plasma parameters and partial differential equations with sparse measurement data.